Methodological discussion of Airbnb's hedonic study A review of the problems and some proposals tested on Bordeaux City data

摘要

Sharing economy-based accommodations has grown dramatically worldwide. Given the crucial role of pricing, academics have used hedonic methods to estimate a large number of implicit prices and accommodation characteristics. It is now time to question our estimation methods. Based on the hedonic methodological literature, this article outlines five proposals to improve model performance: extended specification of attributes, flexible specification of functional forms, market segmentation, and treatment of time and spatial heterogeneities. Each proposal is tested on 13,991 Airbnb listings in Bordeaux, one of the world's top urban tourism destinations. Findings offer significant guidelines for future studies and more relevant information to market players but also suggest that low performance on specific segments is not due to technical problems.